DE4244608C2 - Radar method carried out by means of a computer for measuring distances and relative speeds between a vehicle and obstacles in front of it - Google Patents

Description

The invention relates to a radar method carried out by means of a computer the preamble of the main claim.

So far, radar technology has been used to measure distances and relative speeds primarily used in aviation. There it is about recording one Distance range between a few hundred meters to many kilometers. Against it The special conditions in road traffic require the registration of a Distance range from less than one meter to over a hundred meters. Beyond that Appropriate self-sufficient distance warning and safety systems must be made in view of the usually confusing compared to air routes and complicated street scenes relatively many obstacles at the same time with as many as possible characterizing data. Basically, the capture of three Data, namely distance between vehicle and obstacle, relative speed between vehicle and obstacle (using the Doppler effect) and amplitude of the am Obstacle reflected transmission signal to obtain a statement about the size of the Obstacle desired.

From the German layout specification DE 23 05 941 B2 a vehicle obstacle radar for measuring of distance and relative speed of obstacles known Transmitter quickly changing frequency modulated and unmodulated continuous wave signals via radiates a circulator from an antenna.  

The signals reflected by the obstacles are in two mixers with the transmit mixed signals. The mixers provide the cosine and the sine of the beat frequency. These signals come between to record the relative speed Motor vehicle and obstacle on a speedometer. One of the signals is used for the distance measurement to the obstacle.

In "Kleinheubacher reports" 1991, volume 35, pp 731 to 740 are further continuous wave radar described methods that measure a simultaneous distance and relative speed and allow detection of multiple obstacles. However, series production is one such a radar process as a self-sufficient process from an economic point of view scores are not possible because of high frequency slopes, high difference frequencies arise and the bandwidth to be processed (in the high MHz range) for commercial Signal processors is too large.

Furthermore, a radar system is known from patent application GB 2 249 448 A, at which the transmitted carrier wave in a plurality of stages of different frequencies is divided, the duration of the stages depending on the distance to be detected obstacle is selected. The frequency level of the reflected received signal is then used to obtain in-phase and quadrature signals with the following Frequency level of the transmission signal mixed. These signals are further processed in this way works to determine the distance to the obstacle reflecting the transmission signal can.

The invention is accordingly based on the object of providing a radar method which For several obstacles, the distances, the relative speeds and the Amplitude of the reflected signal used as a measure of the size of the obstacles commercial signal processors allowed to record and thus a series use in Motor vehicle enables.

The object is achieved by the features of patent claim 1. Advantageous training and Further developments are presented in the subclaims.

The method according to the invention includes new modulation steps. The peculiarity the modulation is that the continuous transmission signals in frequency constant Levels of constant length without a time interval are broken down. The temporal  che length of such a stage is advantageously about 20 microseconds; after this Time all signals reflected on the obstacle are detected, i. H. the received The signal of the relevant transmission signal stage is in the steady state. During each frequency constant stage of the reflected received signal, a complex sample that detects the steady phase of the reflection time profile contains. With a step length of 20 µs, this would correspond to a sampling frequency of 50 kHz, which is advantageously relatively low.

According to an embodiment of the invention, the received reflections are processed in the processing tated signals a total of four process steps or measurements are provided. Here The first three measurements are used to measure the obstacles, given by the number the transmission signal stages and the number of distance and relative speed windows, while in the fourth measurement from the pots determined in the first three measurements tical obstacles - the ghost or sham obstacles that arose during the evaluation nisse are separated out, so that only the data for actually existing Obstacles are processed further.

According to a development of the invention, the intersection points become for the comparison ordered in a sequence with decreasing reflection amplitudes. Then it is conjugated complex phase of the highest amplitude intersection with the fourth received signals correlates and the intersection points, the correlation value thus obtained below a pre given value, are excluded as ghost obstacles. In contrast, it becomes iterative in each case after determining an intersection with a high correlation value on the associated Fictitious fourth received signal inferred and this from the received signals the fourth measurement subtracted.

The other subclaims describe calibration methods that are of particular advantage can be used in the context of the inventive method.

An embodiment of the invention is explained below with reference to the drawing. In this show:

Fig. 1 the basic, known structure of a radar apparatus, as it can find also use for the inventive method,

Fig. 2 shows a frequency-time diagram of the method according to the invention,

Fig. 3 shows the frequency-time diagram for distance and Geschwindigkeitsmes solution of an obstacle,

Fig. 4 is a speed-distance diagram of the inventive procedural Rens,

Fig. 5 is an averaging of the invention of the reflected signals,

Fig. 6, a scheme for the overall calibration,

FIGS. 7 to 11, the effect of the calibration on the signal curves and

Fig. 12, 13 and 14 location diagrams without and with phase errors or after the same correction.

Referring first to the block diagram of Fig. 1, so the device is in the radio frequency or microwave part H and the signal processing part S. divided the voltage-controlled oscillator (VCO) in a conventional manner a continuous microwave signal (continuous wave) generated whose frequency For example, by means of a varactor diode over several 100 MHz proportional to a control modulation voltage f (t) can be changed, serves in the method according to the invention for generating frequency-constant, without a time interval between successive transmission signal stages of constant length over the coupler 2 and the circulator 3 , the direction of which is indicated by the arrow, arrive at the transmitting and receiving antenna 4 . 5 with an obstacle on the road in front of the vehicle equipped with this device is designated. The paths of the transmission signal stages and the reflected received signals are indicated by arrows.

The circulator 3 separates the transmitted signals from the received reflected signals, so that only the reflected signals arrive at the right input of the quadrature mixer 6 in the figure, while its left input in the figure is fed out by the coupler 2 . The mixer 6 forms the difference frequency between the transmit and receive signal as an in-phase signal I and a quadrature phase signal Q, both of which are initially present in analog form, ie as sinusoidal time signals. In the system-digital converters 7 and 8 there is a conversion into digital signals y _n , which are fed to the digital signal processor 9 - still to be described in its mode of operation - at the outputs 10 , 11 and 12 of which signals for the respective distance are then obtained Relative speed and the amplitude, ie the obstacle size, can be tapped.

The signal processor 9 , to which the clock generator 13 is assigned, in turn serves via the code generator 14 to generate the control voltage f (t) for the oscillator 1 .

In the following description of the method according to the invention, the requirement is assumed as an example that a distance range of 128 meters should be recorded with a resolution of one meter. A subdivision of the relative speed range to be recorded into 128 windows is also assumed. This results in the need to process 128 × 128 windows, which would not only lead to an unacceptable computing effort, but also to a relatively long computing time, at least above 10 msec. For this reason, the method according to the invention according to FIG. 2, in which the transmission frequency f is plotted over time t, provides three measurements A, B, and C in which the obstacles are detected, and a subsequent fourth measurement D, which serves to sort out false obstacles created in the process.

As already explained, the method according to the invention works with the generation of frequency-constant stages of constant length over time, into which the transmission signals are broken down and which are used in measurements A, B and C. In Fig. 2 is enlarged a sequence of such stages 20 is shown, the duration t _B in the assumed example is 20 microseconds and the frequency _increments f _incr. of 1 MHz. As also already explained, the use of such transmission signal stages 20 enables an evaluation of the reflected signals when the phase is steady.

All four measurements A, B, C and D are carried out sequentially, ie in the device according to FIG. 1 measurement A is carried out first for all 128 stages 20 , then measurement B, and so on. During measurement A, the oscillator 1 is controlled by a corresponding control voltage f (t) in such a way that, starting from a minimum frequency of 77 GHz in this example, a sequence of stages increasing linearly and incrementally to a maximum value of 77, 128 GHz in this example 20 generated. During measurement B, on the other hand, the oscillator 1 , now starting from the maximum frequency, generates a linear sequence of steps with an opposite slope, so that the frequency of 77 GHz is finally reached again. In contrast, during measurement C, 128 stages are transmitted, all of which have the same frequency.

The reflected signals received during the first three measurements A, B and C are each weighted with a Blackman cosine window and input into a Fourier transformation (FFT). As explained in detail in "Introduction to Radar Systems" by MI Skolnik, 1962, page 88, the first two measurements A and B for an obstacle that is distinguished by a high reflection intensity enable the simultaneous measurement of distance and relative speed. In Fig. 3, the curves 30 and 31 of the transmission signals of the received reflected signals are reproduced in a frequency (f) time (t) diagram. L is the echo delay time and Δf _{Doppler is} the Doppler frequency. From this, as well as from the frequency differences Δf _up and Δf _down , the relationships result for the relative speed and the distance or the distance

Δf _Doppler = (Δf _down - Δf _up ) / 2
Δf _Entf = (Δf _down + Δf _up ) / 2.

Measurement C now takes into account the fact that measurements A and B for more than one obstacle due to ambiguity of the mathematical relationships are not useful.

Each Fourier transformation carried out after the three measurements A, B and C with the reflected signals received thereby provides 128 "spectral values" in the assumed example. Outstanding maxima occurring at the output cells of the Fourier transformation are justified by reflections from obstacles. The relationships given above in connection with FIG. 3 apply to the first and second measurements; in measurement C, the frequencies indicated by the maxima are equal to the Doppler frequencies of the obstacles.

In a speed-distance diagram ( FIG. 4), the frequencies determined in the three measurements A, B and C are characterized by throngs of intersecting straight lines, which each represent potential obstacle positions. The straight lines 40 and 41 relate to the first measurement A, the straight lines 42 and 43 to the second measurement B and the straight lines 44 , 45 and 46 to the third measurement C. After the measurement results of the first three measurements have been linked, there are only potential obstacles the intersection of the straight lines 40 to 46 in question. This can be actual obstacles 47 , 48 and 49 or a ghost obstacle 50 that was caused by the link. The first three measurements A, B and C accordingly provide data for a defined number of potential obstacles, which is smaller than the number of original increments, and the fourth measurement D must now be used to sort out the ghost obstacles.

In principle, this is done in that the potential obstacles determined by the measurements A, B and C as intersection points are checked by means of a defined frequency modulation or frequency coding to determine whether there are actual obstacles or ghost obstacles. Thereby, successive stages with frequencies f _{n are created} according to the relationship by means of the oscillator 1

f _n = f _T + f _incr. · (A ⁿ mod (P))
With
n = O. . . N - 1, where N = P - 1 and P = prime number,
A = natural number chosen for the respective length N so that N different coefficients arise;
f _T = carrier frequency of the oscillator,
f _incr. = Frequency increment
mod = modulo operator

is generated, and during each - in each case assigned to one of the coefficients (A ⁿ mod (P)) - reflected stage the fourth received signal becomes a complex sample

with i = number of obstacles,

steady phase value of obstacle i,
c = speed of light,
R _i = distance of the target obstacle i;
v _i = relative speed of the obstacle i,
f _T = carrier frequency of the oscillator,
f _A = sampling frequency,
k _i = amplitude
calculated.

The phases Phasen _{i; n} assigned to the obstacles in the fourth measurement are then compared with the phases of the intersections 47 , 48 , 49 and 50 in the relative speed-distance diagram of FIG. 4.

To facilitate understanding, the following numerical example is assumed for frequency coding f _n :

P = 37; N = P - 1 = 36; A = 5;
f _T = 77 GHz; f _incr. = 1 MHz.

The individual coefficients (A ⁿ mod (P)) are then:
1; 5; 25; 14; 33; 17; 11; 18; 16; 6; 30; 2; 10; 13; 28; 29; 34; 22; 36; 32; 12; 23; 4; 20; 26; 19; 21; 31; 7; 35; 27; 24; 9; 8th; 3; 15.

These coefficients cover all natural numbers between 1 and 36 and are all different from each other. Multiplied by the frequency hopping increment, here 1 MHz, they result in the modulation curve shown in FIG. 2 in area D: For each coefficient one obtains a frequency-constant stage and during each stage a receive sample in the form of I and Q designated in FIG. 1 Signals. Deviating from the stages obtained in the first three measurements A, B and C, however, the stages obtained in the fourth measurement D do not form a linear sequence of stages.

The defined sequence of the coefficients (A ⁿ mod (P)) now has special mathematical properties that form the essential prerequisites for use in the radar system:

• 1. Multiply the sequence by any integer, except 0, and bring the result sequence back into range 1 using modulo calculation. . . N, there is a cyclically shifted version of the episode. If one looks at the example above, the result is when the sequence given there is multiplied by the number 2 and then by mod (37) in area 1 . . . 36 is brought, the episode 2 ; 10 ; 13 ; 28 ; 29 ; . . . which is a cyclically shifted version of the original episode.
• 2. Subtract a cyclically shifted version of this sequence from the coefficient sequence and bring the result back into area 1 by modulo calculation. . . N, another cyclically shifted version of the sequence is created. In the example: From episode 1 ; 5 ; 25 ; 14 ; 33 ; . . . episode 2 ; 10 ; 13 ; 28 ; 29 ; . . . subtracted. This creates the cyclically shifted version 36 ; 32 ; 12 ; 23 ; 4 ; . . .

As can already be seen from the above-mentioned law of formation for the received signal y _{n of} the fourth measurement D, this received signal is composed of the superposition of several reflection processes at obstacles, which are received weighted with different reflection amplitudes k _i and which each have the settled phase values ϕ _{i ; n included} over the discrete time axis n. If one uses the values of the above numerical example in the relationship for ϕ _{i; n} and assumes a sampling frequency of 50 kHz, the result is

The first, distance or distance-related summand is generated thus phases distributed in the complex level. Because the distance enters these phases as a multiplier and through the cyclical periodicity of the phase with respect to 360 degrees a modulo Effect is created according to the above under 1. defi a precondition, depending on the distance, a cyclically shifted Version of the phase sequence.

The potential obstacles obtained by means of the first three measurements A, B and C are now arranged in a sequence with decreasing reflection amplitudes k _i and multiplied by a correlator matching the assumed distance-speed vector in order to sort out the ghost obstacles; then the products are added:

First, the correlation value W for the obstacle with the largest reflection amplitude calculated:

with i = 0 (target with the highest amplitude).

According to this correlation relationship, a complex complex conjugate is multiplied by the received signal y _{n of} the fourth measurement D, which sequence matches the potential obstacle with the strongest reflection amplitude. This means that the corresponding phase sequence is subtracted. A high correlation value W arises only in the case of an actual obstacle, since only then is the received phase sequence y _n reversed by the correlation to a phase-constant sequence. In the case of a ghost obstacle, there is no such phase sequence in the received signal y _n , and the weighting with the correlation generates cyclically shifted versions of the phase sequence in accordance with the mathematical property specified under 2 because of further potential obstacles. The sum of such phase-scattered vectors, ie the associated correlation value W, is low, so that apparent obstacles can be recognized as such and sorted out.

In the diagram according to FIG. 4, the intersection 50 is thus sorted out, and the remaining intersections 47 , 48 and 49 are treated as real obstacles. This is done in such a way that it is iteratively inferred after the determination of an intersection point with a high correlation value W, i.e. here the intersection points 47 , 48 and 49 , that the associated fictitious fourth received signal ^{j ϕ i; n is derived} from the received signal y _{n of} the fourth measurement is subtracted:

for all n = O. . . N - 1.

This difference therefore only contains the reflected signals from the second largest target downwards. As already mentioned, this procedure is repeated iteratively, so that, on the basis of this new received signal y _n , a correlation check is again carried out for the now largest (ie overall second largest) target; this procedure is repeated accordingly.

In short, one can say that for real obstacles, the phase of the received signal assigned to the associated intersection in the diagram according to FIG. 4 is calculated in the manner of an inverse Fourier transformation and compared with the result from the fourth measurement D. Since complex values are available, this check is carried out according to a correlation function; the correlation values only have the result 36 , for example, if there is a real obstacle.

After the described iterative review of all of the measurements solutions A, B and c identified potential obstacles after the fourth measurement D use the actual obstacles high detection reliability available; the false alarm rates regarding bogus obstacles are extremely small.

In principle, it would be possible to use the fourth ver step D distances and relative speeds also at Determine the presence of multiple obstacles. At a sufficiently large resolution would, however, be such a long one Computing time reveals a real-time application of the method would not be possible. For this reason, the first three Process steps A, B and C preselect the potential Hit obstacles, and on the reduced number of fourth Process step D applied. The result is a process that gets along with computers of usual capacity and a safe one Detection of obstacles with a short evaluation time guaranteed animals.

The calibration procedure described below can be used to further increase the accuracy. Since the quadrature mixer 6 contains two mixers for generating a complex signal I, Q, this signal can be essentially falsified by two effects, namely a "offset error" caused by a crosstalk signal and a "quadrature component error".

The generation and compensation of the offset error are described below: The amplitude of the transmission signal fed into the quadrature mixer 6 and thus the response sensitivity of its mixer diodes depends on the respective frequency, set by the control voltage f (t), so that this too Transmitting signal is present directly at the output of the quadrature mixer 6 .

This "crosstalk signal" is compared to the actual one received signal (i.e. the reflected signal) extremely strong, so that the targets or obstacles to be detected, as it were be covered. The level difference between the two signals Depending on the size and distance of the targets, types can be 40. . . 100 dB. The "crosstalk signal" is called an "offset error" referred to as being largely at a certain frequency stationary offset of the output signal on a mixer diode occurs.

The range of fluctuation of the defined offset comprises only one low amplitude range; the fluctuations also have a large time constant (a few minutes) during which the offset superimposed received signal at a considered frequency of Measurement to measurement has a stochastic character. Therefore can the offset error by averaging a large An number of reflected signals at the same frequency can be determined very precisely. This is happening partly during the ongoing operation of the radar system. Is the Well known offset errors, it can be used for any frequency can be corrected simply by the mixer output signal I, Q is subtracted.

The averaging of the received reflected signals of a certain frequency is advantageously not carried out by the arithmetic mean, but by an exponentially decaying past evaluation w (t), which is shown in FIG. 5 over time t (the abscissa value t1 means several minutes) . The mean value calculated by such a past evaluation has three advantages over the arithmetic mean: On the one hand, the received mean value can quickly follow an error drift, since new received reflected signals are weighted more than old ones. Secondly, the mean value is characterized by a low spread, since a large number of signals contribute to the average. Third, such a past assessment is easy to implement; the last calculated mean at a certain frequency f is renewed at the time of the return of this frequency f by a simple recursive rule. This rule is explained below.

The current mean offset value (Re = real part, Im = imaginary part) at frequency f is calculated to a large extent from the last mean offset value at the same frequency, added to a small part of the new sample value Re _new (f), Im _new (f):

In Fig. 6 is a scheme for the overall calibration Darge provides. The real and imaginary components are given for a frequency f. In the left part of this figure, the offset correction 0 of the signal arriving from the mixer diodes is symbolized by subtracting the mean offset values.

Figs. 7 to 11 show the effect of this calibration: FIG. 7 again shows the temporal frequency response in the measurements A to D of Figure 2. From a stationary obstacle, for example, 4 m spacing of the quadrature mixer supplies 6 (see Fig. 1.. ) without calibration, the signal curves according to FIGS. 8 and 9, on the other hand with the described calibration method according to FIGS. 10 and 11.

The average offset values can be updated by the recursive calculation rule whenever a certain frequency value (level) returns. However, the duration of the return is different in the selected modulation characteristic according to FIG. 2. Therefore, the entire measurement time for the four individual measurements AD (see FIG. 2) is selected as the update period T _AK for the evaluation of a radar image, so that all frequencies are treated under the same conditions. The constant Z must correspond to the desired time constant τ _corr. be selected for the exponentially decaying past evaluation:

If, for example, the total measuring time of the radar evaluation is 10 ms, the value Z = 0.001 results with a recursion time constant of τ _corr = 10 s; ie the current mean offset error is calculated to be 99.9% from the old mean offset error added to 0.1% of the current received signal.

The quadrature component error of the quadrature mixer 6 is characterized by an amplitude and a phase error component.

If one first considers the amplitude error, it is characterized by different response sensitivities of the mixer diodes of the quadrature mixer 6 with the same RF supply power. To correct the amplitude error, the ratio B (f) is important, which represents the quotient from the mean values of the real part and imaginary part after offset correction. Since the quantity B (f) is frequency-dependent, it is determined for every frequency in the receiving operation of the radar system, as with offset correction:

Both mean values

(Mean values of Amounts) are calculated using the method of ex paston value decaying is calculated.

To correct the amplitude error, the determined factor B (f) is multiplied by the imaginary part Im (f) (see FIG. 6).

After the amplitude error of the quadrature components has been corrected, the phase error remains. It indicates the deviation from the phase value 90 ° of the two signals to be fed into the mixer. For the effect of such a phase error, the illustration in FIG. 12 should be considered:

In FIG. 12, approximately 10,000 samples of a received signal assumed as an example are entered in location diagrams. In this case, only samples with regard to a certain recurring frequency were selected. Provided that the target situation is not so stationary that distances change by less than a fraction of an RF wavelength (approx. 4 mm) within a few minutes, a larger number of received signals always have a stochastic character. The stochastic character is evident in the location diagram in the form of a two-dimensional Gaussian cloud; ie any cross-section of such a cloud through the coordinate origin creates a one-dimensional view of a Gaussian distribution density function.

The diagram of FIG. 12 shows a concentric "cloud" as it looks without phase and amplitude errors at the output of the quadrature mixer 6 . The diagram in Figure 13 shows the result of a faulty quadrature mixer. After long-term observation, one can see the elliptical shape of the cloud with any directional axis, since there are amplitude and phase errors. After the amplitude correction described above, the phase errors remain; the elliptical expression is now only available in the 45-degree direction (see Fig. 14).

Since the directional axis of the ellipse is definitely in the 45-degree direction, all reception values are rotated by 45 degrees to eliminate this phase error, so that the elliptical directional axis is now in the 90-degree direction. As shown in FIG. 6, this rotation takes place by crosswise addition or subtraction of the quadrature components. Although the amplitude of the signals is additionally increased by √, this change in amplitude does not have a disruptive effect, but is only included in the processing as a standardization constant.

After the 45 ° rotation of the ellipse, the length-to-side ratio can be determined by the quotient ψ (f), which is obtained by dividing the mean value of the real part by the mean value of the imaginary part. However, this corresponds to the same procedure as for amplitude correction, since the phase errors of the quadrature components have become amplitude errors due to the 45-degree rotation. The phase error is therefore also corrected in accordance with the amplitude correction by multiplying the imaginary part by the factor ψ (f) (see FIG. 6).

Figure 6 shows all the corrective measures for the signal y _n coming from the A / D converters 7 and 8 up to the final, almost ideal received signal Re _corr (f), Im _corr (f). The question arises as to whether and to what extent the individual corrective measures influence each other or even lead to "control vibrations".

The amplitude correction and the phase correction of the quadrature components do not influence each other, because they interact which are orthogonal.

The offset correction and the correction of the quadrature components can interfere with each other, because z. B. an unbalanced offset error can lead to a falsification in the amplitude correction. Therefore, the time constant τ _corr in the offset correction is designed to be approximately 10 times lower than in the correction of the quadrature component errors.

Claims (15)

1. Radar method carried out by means of a computer for measuring distances and relative speeds between a vehicle and obstacles in front of it, with continuous transmission signals generated by means of an oscillator, during the transmission of the continuous transmission signals, simultaneous reception of signals reflected at the obstacles, mixing of the reflected signals with the continuous transmission signals for obtaining in-phase and quadrature signals and processing these signals into output signals for the distances and relative speeds of the obstacles, characterized in that the continuous transmission signals are broken down into frequency-constant stages of constant length without a time interval and that during each frequency-constant stage a complex sample value of the reflected received signal is recorded and mixed with the transmitted signal of the same frequency-constant stage. 2. The method according to claim 1, characterized in that the length of the frequency constant levels is in the range of 20 µs.   3. The method according to claim 1 or 2, characterized in that during a first measurement of the oscillator in the sense of generating frequency constant, temporally without spacing successive stages with in this sequence from a minimum to a maximum value linearly increasing in frequency and with the desired Resolution given number is driven that is then driven during a second measurement solution of the oscillator in the sense of generating corresponding frequency-constant stages with frequency decreasing linearly from the maximum value to the minimum value, with a complex in both measurements during each frequency-constant stage of the received reflected signal Sampled value acquired and by mixing with the transmission signal of the same frequency-constant stage first or second in-phase and quadrature phase signals (first or second reception signals) for the distances and the relative speeds are obtained that would then nd a third measurement of the oscillator is driven in the sense of the generation of corresponding, but frequency-equal, frequency-constant stages, and here, too, a complex sample value for obtaining third in-phase and quadrature-phase signals (third reception signals) for the relative speeds by each stage of the received, reflected signal Mixing with the transmission signal of the same frequency constant stage is detected that all received signals are converted by means of Fourier transformation into relative speed and distance-dependent frequency values which, in a relative speed-distance diagram, represent three sets of intersecting straight lines whose intersections represent potential obstacles, that furthermore, by eliminating ghost obstacles by correlating sample values during a fourth measurement, the oscillator in the sense of generating temporally on successive stages with frequencies f _n according to the relationship hung f _n = f _T + f _incr. (A ⁿ mod (P))
With
n = 0. . . N - 1, where N = P - 1 and P = prime number,
A = natural number chosen for the respective length N so that N different coefficients arise,
f _T = carrier frequency of the oscillator,
f _incr. = Frequency increment, is controlled, furthermore, during each - in each case assigned to one of the coefficients (A ⁿ mod (P)) - a complex sample value as a fourth received signal With
i = number of obstacles, c = speed of light,
R _i = distance of the obstacle i;
v _i = relative speed of the obstacle i,
f _T = carrier frequency,
f _A = sampling frequency,
k _i = amplitude is detected and a comparison of the sample values assigned to the obstacles in the fourth measurement in the form of phases (ϕ _{i; n} ) is carried out with the sample values of the intersection points in the phases in the relative speed-distance diagram. 4. The method according to claim 3, characterized in that for the comparison, the intersection points are arranged in a sequence with decreasing reflection amplitudes (k _i ), the conjugated complex phase of the amplitude-largest intersection point with the fourth received signals (y _n ) according to the function is correlated, intersections whose correlation value (W) thus obtained is below a predetermined value are sorted out as ghost obstacles, on the other hand iteratively each time after determining an intersection with a high correlation value (W) on the associated fictitious fourth received signal · e ^{j ϕ i; n} inferred and this is subtracted from the received signals (y _n ) of the fourth measurement. 5. The method according to any one of claims 1 to 4, characterized in that to eliminate transmission signals contained in the in-phase and quadrature phase signals (I, Q) ("crosstalk signal") first by averaging received signals received by all of the control voltage ( f (t)) of the oscillator ( 1 ) given frequencies of the same, the crosstalk signal is determined and then subtracted from the in-phase and quadrature phase signals (I, Q). 6. The method according to claim 5, characterized in that a moving averaging with exponential over time sounding past evaluation (w (t)). 7. The method according to claim 5 or 6, characterized in that that updated new crosstalk signals,  at a given frequency recursively from one previous over determined by averaging speech signal, and a portion of a new crosstalk signal will. 8. The method according to claim 3 or 4 and one of claims 5 to 7, characterized in that the determination of the over speech signals at each burst with the specified frequency he follows. 9. The method according to claim 3 or 4 and claims 7 and 8, characterized in that the update period measurement time required for the four measurements (A, B, C, D) is selected. 10. The method according to any one of claims 1 to 9, characterized records that to correct a quadrature component error in the in-phase and quadra obtained by mixing turphase signals (I, Q) first an amplitude error component  this error by averaging the amounts of these two signals (I, Q) - if necessary after eliminating the Crosstalk error - and then a phase error component of the Quadrature error due to phase rotation of in-phase and Quadrature phase signals (I, Q) is eliminated. 11. The method according to claim 10, characterized in that a moving averaging with exponential over time sounding past evaluation (w (t)). 12. The method according to claim 10 or 11, characterized in that updated new amount values at a given one Frequency recursively from averaging Amount values and a portion of the amount of a new one In-phase and quadrature phase signals (I, Q) - if necessary according to Elimi nation of the crosstalk error - can be determined. 13. The method according to any one of claims 10 to 12, characterized ge indicates that the quotient of the average values of in-phase and quadrature phase signals (I, Q) with the Quadrature phase signal (Q) - if necessary after eliminating the Crosstalk error - is multiplied. 14. The method according to any one of claims 10 to 13, characterized ge indicates that to eliminate the phase error component a phase rotation of 45 ° by crosswise addition or Subtraction between the in-phase and quadrature phase signals takes place at which there is a multiplication of the quadrature phase signals (Q) with the quotient of the mean values of in-phase and quadrature phase signals (I, Q). 15. The method according to claims 5 to 14 with elimination of Crosstalk signal and correction of the quadrature components error, characterized in that the correction with time constant smaller than the elimination.